(AIM)— In a groundbreaking study, Google DeepMind has utilized artificial intelligence (AI) to predict the structures of over two million new materials, an achievement that could soon be applied to improving real-world technology. In a research paper published in the journal “Nature,” the AI company under Alphabet stated that nearly 400,000 hypothetical material designs could soon be manufactured under laboratory conditions.
The discovery and synthesis of new materials are typically a time-consuming and costly process. For example, the research on lithium-ion batteries, now widely used as a power source for everything from smartphones and laptops to electric cars, took approximately 20 years. Ekin Dogus Cubuk, a research scientist at DeepMind, stated, “We hope to significantly shorten this timeline of 10 to 20 years to a more manageable level through significant improvements in experimentation, automated synthesis, and machine learning models.”
DeepMind’s AI is trained on data from the Materials Project, an international research team established at Lawrence Berkeley National Laboratory in 2011, which has collected existing research data on about 50,000 known materials. Kristin Persson, the director of the Materials Project, said, “We need to create new materials to address global environmental and climate challenges. Through materials innovation, we can develop recyclable plastics, harness waste energy, manufacture better-performing batteries, and build cheaper, longer-lasting solar panels, among other things.”
Google DeepMind has developed a deep learning tool called Graph Networks for Materials Exploration (GNoME) for generating new data. Researchers used the workflow and data developed over a decade by the Materials Project to train GNoME, and they improved the GNoME algorithm through active learning. GNoME researchers ultimately produced 2.2 million crystal structures, including 380,000 structures they consider stable, which are being added to the Materials Project. These materials are expected to have potential uses in future technology.
Cubuk stated, “We hope the GNoME project can drive research in inorganic crystals.” He also pointed out that external researchers have independently verified more than 700 new materials discovered by GNoME through parallel physical experiments, confirming that their model’s discoveries can be realized in the laboratory.
Persson expressed great excitement about this: “I am thrilled that people are using our work to generate an unprecedented amount of materials information. This is exactly why I set up the Materials Project: to not only make the data I generate freely available to accelerate global materials design but also to show the world what computation can do for you. It can scan a wide range of new compounds and properties more efficiently and rapidly than individual experiments.”
This research highlights the enormous potential of AI in accelerating the discovery and synthesis of new materials. The advancements in this technology could accelerate the development of many industries, particularly in the energy and technology sectors. This achievement underscores the potential role of AI in materials science, with immeasurable contributions to future technological advancements.
For more insights and updates on AI, follow us on:
More information on AI can be found at AI INSIGHT MEDIA (AIM) www.aiinsightmedia.com.
Keywords: Google DeepMind, artificial intelligence, new materials, Materials Project, deep learning, GNoME, crystal structures, research, technology, materials science.
